Research On Ambiguity Elimination And Channel Order Identification In Blind Channel Estimation For MIMO Systems

Being a vital technology in multi-antenna systems,channel estimation has a significant influence on signal detection.Different from conventional channel estimation,blind channel estimation hardly needs training sequences.Then,it can increase the spectrum efficiency,and reduce the requirement of cooperativity between the transmitter and the receiver.Hence,blind channel estimation has become a research focus recently.Focusing on the existence of inherent ambiguity of blind channel estimation and the requirement of channel order,we investigate two key technologies,ambiguity elimination and channel order estimation,in blind channel estimation for MIMO systems.The major contributions of this work are summarized as follows:For the scalar ambiguity of the subspace-based blind channel estimation methods for MIMO systems,we investigate the permutation ambiguity caused by multiple transmit antennas and the phase ambiguity caused by the rotation symmetry of constellations,with the assumption of known channel order.Furthermore,we propose an ambiguity-elimination based blind channel estimation method exploiting training sequences.In the proposed method,we design a timing orthogonal training sequence to eliminate the permutation ambiguity,and utilize two order-coprime modulations to eliminate the phase ambiguity.As the serious effect on blind channel estimation of the inaccurate result of channel order estimation,the research on the methods for channel order estimation is necessary.Conventional channel order estimation methods perform not well at a low signal-to-noise ratio(SNR)for a small sample size,therefore,we propose two channel order estimation methods,which are Wishart matrix based and Gerschgorin disks based,respectively.The Wishart matrix based method utilizes the distribution property of noise eigenvalues to distinguish the signal subspace and the noise subspace.The Gerschgorin disks based method utilizes the radii of Gerschgorin disks calculated by the covariance matrix after unitary transformation to derive an adaptive threshold which is used to distinguish the two subspaces.The proposed subspace-based blind channel estimation method exploiting training sequence can effectively eliminate the scalar ambiguity,which is verified by simulation results.Besides,the two channel order estimation methods can achieve a reliable estimation at a low SNR for a small sample size,which is also verified by simulation results.